marcspec 0.2.1 → 0.5.0

data/CHANGES

@@ -0,0 +1,15 @@
+0.5.0
+  * Allow solr field names to repeat in a spec set. This allows you to cumulatively add to a solr field based on "regular" 
+    and custom (or, say, two custom) specs.
+0.4.0
+  * MAJOR BACKWARD-INCOMPATIBLE CHANGE!! The signature for custom routines is now def function(doc, record, my, args), where
+  "doc" is a hashlike (usually a SolrDocument) that contains all the work that has happened up to this point. The idea is that you 
+  can use previously-computed values to determine values for different fields. Use sparingly.
+0.3.0
+  * Changed behavior with respect to repeated subfields. A spec such as '260 ac' returns the values of the 'a' and
+  'c' subfields concatenated together. A request for '631 z' would, similarly, return the values *of all the subfield
+  z's* concatenated together. I'm now treating this as a special case. If the request is for a single subfield code,
+  multiple values are returned separate from each other.
+  
+0.2.0
+  * First public release

data/README.rdoc

@@ -21,7 +21,7 @@ options.
 * A MARC Field Spec is one of the following
   * A {MARCSpec::LeaderSpec} for dealing with the leader
   * A {MARCSpec::ControlFieldSpec}, consisting of a tag (as a string, not a number) and
-    an optional zero-based index (e.g., "001[3]") or range (e.g., "001\[11..13\]")
+    an optional zero-based index (e.g., "001[3]") or range (e.g., "001[11..13]")
   * A {MARCSpec::VariableFieldSpec}, consisting of a tag, a couple indicator patterns
     (currently ignored, but stay tuned), an optional list of subfields (default
     is all), and an optional string used to join the subfields (default is

data/VERSION

@@ -1 +1 @@
-0.2.1
+0.5.0

data/lib/marcspec/customspec.rb

@@ -49,8 +49,9 @@ module MARCSpec
       
     end
     
-    def raw_marc_values r
-      return @module.send(@methodSymbol, r, *@methodArgs)
+    
+    def raw_marc_values r, doc
+      return @module.send(@methodSymbol, doc, r, *@methodArgs)
     end
     
     def self.fromHash h

data/lib/marcspec/marcfieldspec.rb

@@ -176,8 +176,10 @@ module MARCSpec
       vals = []
       fields.each do |f|
         subvals = f.sub_values(@codes)
-        vals << subvals.join(@joiner) if subvals.size > 0
+        subvals =  subvals.join(@joiner) if subvals.size > 0 and (@codes.nil? or @codes.size > 1)
+        vals << subvals
       end
+      vals.flatten!
       return vals
     end
 

data/lib/marcspec/solrfieldspec.rb

@@ -17,7 +17,7 @@ module MARCSpec
     end
 
 
-    def raw_marc_values r
+    def raw_marc_values r, doc
       vals = []
       @marcfieldspecs.each do |ts|
         vals.concat ts.marc_values(r)
@@ -25,8 +25,8 @@ module MARCSpec
       return vals
     end
       
-    def marc_values r
-      vals = raw_marc_values r
+    def marc_values r, doc = {}
+      vals = raw_marc_values r, doc
       unless vals.is_a? Array
         vals = [vals]
       end

data/lib/marcspec/specset.rb

@@ -1,5 +1,23 @@
 require 'jruby_streaming_update_solr_server'
+
+
+    
+
 module MARCSpec
+
+  # Create a mock solr document based on a normal hash for mocking.
+  # All we really need is a compatible add method
+  class MockSolrDoc < Hash
+    def add key, value
+      if self.has_key? key
+        self[key] << value
+      else
+        self[key] = [value]
+      end
+      self[key].flatten!      
+    end
+  end
+
   class SpecSet
     attr_accessor :tmaps, :solrfieldspecs
     
@@ -21,10 +39,35 @@ module MARCSpec
       end
     end
     
+  
+    
     def add_map map
       self.tmaps[map.mapname] = map
     end
     
+    
+    def buildSpecsFromList speclist
+      speclist.each do |spechash|
+        if spechash[:module]
+          solrspec = MARCSpec::CustomSolrSpec.fromHash(spechash)
+        else
+          solrspec = MARCSpec::SolrFieldSpec.fromHash(spechash)
+        end
+        if spechash[:mapname]
+          map = self.map(spechash[:mapname])
+          unless map
+            $LOG.error "  Cannot find map #{spechash[:mapname]} for field #{spechash[:solrField]}"
+          else
+            $LOG.debug "  Found map #{spechash[:mapname]} for field #{spechash[:solrField]}"
+            solrspec.map = map
+          end
+        end
+        self.add_spec solrspec
+        $LOG.debug "Added spec #{solrspec.solrField}"
+      end
+    end
+      
+    
     def add_spec solrfieldspec
       self.solrfieldspecs << solrfieldspec
     end
@@ -39,7 +82,8 @@ module MARCSpec
     
     def fill_hashlike_from_marc r, hashlike
       @solrfieldspecs.each do |sfs|
-        hashlike[sfs.solrField] = sfs.marc_values(r)
+        hashlike.add(sfs.solrField,sfs.marc_values(r, hashlike))
+        # hashlike[sfs.solrField] = sfs.marc_values(r, hashlike)
       end
     end
 
@@ -50,7 +94,7 @@ module MARCSpec
     end      
     
     def hash_from_marc r
-      h = {}
+      h = MARCSpec::MockSolrDoc.new
       fill_hashlike_from_marc r, h
       return h
     end

data/spec/marcfieldspecs_spec.rb

@@ -84,6 +84,11 @@ describe "VariableFieldSpec" do
     dfs.marc_values(@one).should.equal ["Medina, Texas,"]
   end
   
+  it "should return separate values for repeated subfields if only one code is specified" do
+    dfs = MARCSpec::VariableFieldSpec.new('651', 'z')
+    dfs.marc_values(@one).sort.should.equal ['Texas', 'United States of America.']
+  end
+  
   it "Should get all fields via several equal routes" do
     a = MARCSpec::VariableFieldSpec.new('260').marc_values(@one)
     ac =  MARCSpec::VariableFieldSpec.new('260', ['a', 'c']).marc_values(@one)

data/spec/solrfieldspec_spec.rb

@@ -126,7 +126,7 @@ end
 
 module A
   module B
-    def self.titleUp r, codes=nil
+    def self.titleUp doc, r, codes=nil
       title = r['245']
       if codes
         return [title.sub_values(codes).join(' ').upcase]
@@ -134,6 +134,18 @@ module A
         return [title.value.upcase]
       end
     end
+    
+    # downcase and strip punctuation to create a sortable entitiy. Here, we're actually
+    # ignoring the passed-in record entirely
+    def self.sortable doc, r, whereTheTitleIs
+      vals = []
+      if doc[]
+        doc[whereTheTitleIs].each do |title|
+          vals << title.gsub(/\p{Punct}/, ' ').gsub(/\s+/, ' ').strip.downcase
+        end
+      end
+      return vals
+    end
   end
 end
 
@@ -173,5 +185,12 @@ describe "CustomSolrSpec" do
     css.marc_values(@one).should.equal [@mapValue]
   end
   
+  it "works with a map that has multiple return values" do 
+    @map[@titleACValue.upcase] = ['two', 'one']
+    css = MARCSpec::CustomSolrSpec.new(:solrField=>'solrField', :map=>@map, :module=>A::B, :methodSymbol=>:titleUp, :methodArgs=>[['a', 'c']])
+    css.marc_values(@one).should.equal ['two', 'one']
+  end
+  
+  
 end
   

data/spec/specset_spec.rb

@@ -0,0 +1,88 @@
+require 'spec_helper'
+
+module A
+  module B
+    def self.titleUp doc, r, codes=nil
+      title = r['245']
+      if codes
+        return [title.sub_values(codes).join(' ').upcase]
+      else  
+        return [title.value.upcase]
+      end
+    end
+    
+    # downcase and strip punctuation to create a sortable entitiy. Here, we're actually
+    # ignoring the passed-in record entirely
+    def self.sortable doc, r, whereTheTitleIs
+      vals = []
+      if doc[whereTheTitleIs]
+        doc[whereTheTitleIs].each do |title|
+          vals << title.gsub(/\p{Punct}/, ' ').gsub(/\s+/, ' ').strip.downcase
+        end
+      end
+      return vals
+    end
+  end
+end
+
+
+describe "SpecSet Basics" do
+  before do
+    @one = MARC4J4R::Reader.new("#{DIR}/data/one.dat").first
+    @places = ['Medina', 'Texas', 'United States of America.', 'Medina, Texas,']
+    @title = ['The Texas ranger [sound recording] / Sung by Beale D. Taylor.']
+    @titleA = ['The Texas ranger']
+    @speclist = [
+      {
+       :solrField=> "places",
+       :specs => [
+        ["260", "*", "*", "a"],
+        ["651", "*", "*", "a"],
+        ["651", "*", "*", "z"],
+       ]
+      },
+     {:solrField=>'title', :specs=>[['245', '*', '*']]},
+      {
+        :solrField => 'titleA',
+        :specs => [['245', '*', '*', 'a']]
+      }
+    ]
+  end
+  
+  it "should build from a list" do 
+    ss = MARCSpec::SpecSet.new
+    ss.buildSpecsFromList(@speclist)
+    ss.solrfieldspecs.size.should.equal 3
+    h = ss.hash_from_marc @one
+    h['places'].sort.should.equal @places.sort
+    h['title'].should.equal @title
+    h['titleA'].should.equal @titleA
+  end
+  
+  it "allows customs that reference previous work" do
+    @speclist << {:solrField=>'titleSort', :module=>A::B, :methodSymbol=>:sortable, :methodArgs=>['title']}
+    ss = MARCSpec::SpecSet.new
+    ss.buildSpecsFromList(@speclist)
+    h = ss.hash_from_marc @one
+    h['title'].should.equal @title
+    h['titleSort'].should.equal @title.map{|a| a.gsub(/\p{Punct}/, ' ').gsub(/\s+/, ' ').strip.downcase}
+  end
+  
+  it "should allow repeated solrFields" do
+    @speclist << {:solrField=>'titleA', :specs=>[['260', '*', '*', 'c']]} # '1939.'
+    expected = @titleA
+    expected << '1939.'
+    ss = MARCSpec::SpecSet.new
+    ss.buildSpecsFromList(@speclist)
+    h = ss.hash_from_marc @one
+    h['titleA'].sort.should.equal expected.sort
+  end
+end
+   
+
+
+
+
+
+  
+    

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: marcspec
 version: !ruby/object:Gem::Version 
-  hash: 21
+  hash: 11
   prerelease: false
   segments: 
   - 0
-  - 2
-  - 1
-  version: 0.2.1
+  - 5
+  - 0
+  version: 0.5.0
 platform: ruby
 authors: 
 - BillDueber
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-08-11 00:00:00 -04:00
+date: 2010-08-16 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -90,6 +90,7 @@ extra_rdoc_files:
 files: 
 - .document
 - .gitignore
+- CHANGES
 - LICENSE
 - README.rdoc
 - Rakefile
@@ -124,6 +125,7 @@ files:
 - spec/marcspec_spec.rb
 - spec/solrfieldspec_spec.rb
 - spec/spec_helper.rb
+- spec/specset_spec.rb
 has_rdoc: true
 homepage: http://github.com/billdueber/marcspec
 licenses: []
@@ -164,3 +166,4 @@ test_files:
 - spec/marcspec_spec.rb
 - spec/solrfieldspec_spec.rb
 - spec/spec_helper.rb
+- spec/specset_spec.rb